Tools & Products

ChatGPT Launches Virtual Try-On Worldwide: Real-Time Garment Fitting on Personal Photos

OpenAI deployed Virtual Try-On globally across ChatGPT iOS, Android, and web clients, enabling consumers to preview apparel catalog items mapped directly onto uploaded full-body photographs with photorealistic draping and texture preservation.

By FreakVinci · 2026-10-03 · 12 min read

OpenAI expanded ChatGPT's shopping capabilities with the global deployment of Virtual Try-On. The feature allows users to inspect clothes discovered in shopping queries and visualize how garments drape across their own bodies using an uploaded personal photo.

The rollout marks OpenAI's clearest bid to capture high-intent commercial search traffic from Google Shopping and Amazon, turning conversational product research into an interactive fitting room.

Technical Architecture: From Product Grid to Garment Warp

Previous attempts at consumer virtual try-on suffered from unnatural warping, misaligned fabric textures, and distorted limbs. OpenAI implements a three-stage neural pipeline:

┌────────────────────────┐         ┌────────────────────────┐
│  User Reference Photo  │         │  Catalog Product Image │
│  (Posture & Lighting)  │         │  (Flat Lay / Mannequin)│
└───────────┬────────────┘         └───────────┬────────────┘
            │                                  │
            ▼                                  ▼
┌────────────────────────┐         ┌────────────────────────┐
│ 1. DensePose & Normal  │         │ 2. Garment Segmentation│
│    Estimation Model    │         │    & Texture Extraction│
└───────────┬────────────┘         └───────────┬────────────┘
            │                                  │
            └─────────────────┬────────────────┘
                              ▼
┌───────────────────────────────────────────────────────────┐
│ 3. Physics-Informed Latent Inpainting Network             │
│    • Fabric drape & gravity simulation                    │
│    • Specular lighting & shadow harmonizing               │
│    • Photorealistic seam preservation                     │
└─────────────────────────────┬─────────────────────────────┘
                              ▼
                [Fitted Output Preview]
  1. Body Pose and Depth Extraction: The model scans the user's reference photograph using surface normal estimations and dense pose keypoints, identifying torso rotation, arm positions, and shoulder width.
  2. Garment Isolation: The target clothing item is segmented from retail product cards, isolating fabric weave, collars, cuffs, and prints from white catalog backgrounds.
  3. Lighting and Crease Harmonization: A fine-tuned diffusion model synthesizes realistic folds, stretch points, and cast shadows that match the light direction and intensity of the user's uploaded room setting.

Using Virtual Try-On: Step-by-Step

Testing the feature in ChatGPT requires four simple steps:

  1. Prompt for Products: Search for an apparel query, such as "Show me dark navy wool overcoats under $250".
  2. Tap "Try On": When a shopping card appears, click the Try On icon on the bottom corner of the product image.
  3. Select or Upload Photo: Select your stored reference photo from the secure user vault, or upload a new well-lit full-body image.
  4. Inspect Fitting and Angles: The engine generates the composite image in roughly 5.8 seconds, displaying zoomable previews alongside direct checkout links.

Privacy and Biometric Protections

Consumer advocacy groups have scrutinized AI camera features regarding body data harvesting. OpenAI instituted strict operational constraints around reference images:

Privacy Guardrail Implementation
Photo Retention Policy Photos stored in encrypted user-isolated storage; deleted instantly upon user request.
Model Retraining Exemption Reference photos and fitting composites are excluded from OpenAI foundation model pre-training.
Facial Anonymization Toggle Optional automated face-blurring filter runs locally in browser before upload.
Minor Protection Filter Computer vision safeguards reject photos identified as individuals under 18 years of age.

Commercial Impact on Retail Returns

Online clothing retailers battle return rates averaging 24% to 38%, with 65% of returns caused by poor sizing and unflattering drape.

Alpha testing data shared with partner retailers reported an 18% increase in cart conversion rates and a 29% reduction in fit-related item returns among users who previewed purchases through Virtual Try-On before checking out.